AbstractThis paper addresses the design of predictive

networked controller with adaptation of a communication delay. The

networked control system contains random delays from sensor to
controller and from controller to actuator. The proposed predictive
controller includes an adaptation loop which decreases the influence
of communication delay on the control performance. Also, the
predictive controller contains a filter which improves the robustness
of the control system. The performance of the proposed adaptive
predictive controller is demonstrated by simulation results in
comparison with PI controller and predictive controller with constant
delay.

KeywordsPredictive control, adaptation, communication delay,
communication network.
I. INTRODUCTION
IDE class of industrial processes has a similar behavior
that can be described by the mathematical model with
time delay. These are thermal and chemical processes,
pneumatic systems with long transmission, transportation
systems, etc. The time delay may causes unstability and/or
degradation of performance of controlled process when
exceeds a critical value. The traditional PID controller is used
for controlling of processes when time delay has a small
value. In the case of large value of time delay, PID controllers
exhibit poor control performance [1].
Many modern industrial systems are hierarchical organized
and distributed over the communication network with
decentralized control [2]. A feedback system wherein the
control loop is closed through a real-time communication
network is known as a networked control system (NCS). The
main feature of NCS is the exchange, through communication
networks, of system information and control signals between
various physical components. So this type of system has the
advantage of greater flexibility over traditional control
systems, including greater flexibility in diagnosis and
maintenance procedures.
The control over networks is the next frontier. Control Area
Network (CAN) has the potential of supporting interactions
between sensors, actuators, and even micro-controllers


Manuscript received April 23, 2008.
J. Velagic is with the Faculty of Electrical Engineering, Department of
Automatic Control and Electronics, Sarajevo, Bosnia and Herzegovina
(phone: +387 33 25 07 65; fax: +387 33 25 07 25; e-mail:
jasmin.velagic@etf.unsa.ba).
embedded in a plant, all without physical connections. With
appropriate software, embedded devices may be able to
automatically connect to each other, form control loops, and
even self-assemble into fully functional applications on-the-
fly. Currently, a typical mid-range automobile has about 45
micro-controllers connected by CAN [3]. There are also
several challenges though. For example, one has to contend
with unreliable communication and delays. Also, the overall
system features the integration of several complex
technologies such as networking, sensing, actuation,
computing, and control. In addition, there are the difficulties
of dealing with a distributed system. Finally, one must design
such systems to be reliable [4].
The overall NCS performance is always affected by
network delays since the network is tied with the control
system. Delays are widely known to degrade the performance
of a control system. Existing constant time-delay control
methodologies may not be directly suitable for controlling a
system over the network since network delays are usually
time-varying, especially in the Internet. Therefore, to handle
network delays in a closed-loop control system over a
network, an advanced methodology is required [5].
A number of time delay compensation and prediction
schemes have been developed and/or improved with
modifications as shown in [6] and [7]. The performance of
Smith Predictor Control was studied experimentally in [8]. It
shows that the system performs well if the process model is
accurate, but that performance degrades rapidly with
inaccuracy in the process parameters and time delay.
Several control design methods for systems with varying
time delays have appeared in recent literature including an
estimation and self-tuning method proposed in [9], a variable
structure controller [6], and a model reference adaptive
approach [10]. Also, many control methodologies, such as
probabilistic predictor-based delay compensation
methodology [11], optimal stochastic control methodology
(LQG) [12], non-linear and perturbation theory [13], sampling
time scheduling methodology [14] and fuzzy logic modulation
methodology [15], have developed for networked control
system. For systems with large time delays, most design
approaches use a prediction mechanism as part of the
controller to simulate the process for given system parameters
and time delay.
In this paper we proposed a predictive controller with
adaptation of variable time delay between sensor and
Design of Smith-like Predictive Controller with
Communication Delay Adaptation
Jasmin Velagic
W
World Academy of Science, Engineering and Technology
Vol:2 2008-11-22
172
I
n
t
e
r
n
a
t
i
o
n
a
l

S
c
i
e
n
c
e

I
n
d
e
x

V
o
l
:
2
,

N
o
:
1
1
,

2
0
0
8

w
a
s
e
t
.
o
r
g
/
P
u
b
l
i
c
a
t
i
o
n
/
1
4
0
0
1



controller with aim to decrease an influence of communication
delays on the closed-loop control performance over the CAN
network.
II. NETWORKED PREDICTIVE CONTROL SYSTEM
The proposed predictive control system with adaptation of
communication delay is shown in Fig. 1. This system includes
PI controller, DC motor as a plant, discrete model of process,
communication network, adaptation loop of communication
delay, filter and appropriate sensor and actuator. Controller,
sensor and actuator represent communication network nodes.
The brief description of proposed control system is given in
this section.


Controller Plant
Model of
process
Adaptation
of delay
Filter
Adaptive Predictor
Actuator Sensor
Network
r e y
yp
ep
ya
ym
u


Fig. 1 Predictive Networked Control System with Adaptation of
Communication Delay

A. Description of Communication Network
In this paper we used CAN communication network under
TrueTime Matlab Toolbox
1
. This toolbox simulates medium
access and packet transmission in the CAN network and
supports CSMA/AMP protocol. It stands for Carrier Sense
Multiple Access with Arbitration on Message Priority. If the
network is busy, the sender will wait until it occurs to be free.
If a collision occurs (again, if two transmissions are being
started within 1 microsecond), the message with the highest
priority (the lowest priority number) will continue to be
transmitted. If two messages with the same priority seek
transmission simultaneously, an arbitrary choice is made as to
which is transmitted first. (In real CAN applications, all
sending nodes have a unique identifier, which serves as the
message priority).
Controller, sensor and actuator are connected through
network. Data transfers between the controller and the
actuator will induce network delays in addition to the
controller processing delay. Network delays in an NCS can be
categorized from the direction of data transfers as the sensor-
to-controller delay
sc
and the controller-to-actuator delay
ca
.
The delays are computed as:


1
TrueTime Toolbox is product of the Department of Automatic Control,
Lund Institute of Technology, Sweden.

,
,
ce rs ca
se cs sc
t t
t t
=
=

, (1)

where t
se
is the time instant that the actuator encapsulates the
measurement to a frame or a packet to be sent, t
cs
is the time
instant that the controller starts processing the measurement in
the delivered frame or packet, t
ce
is the time instant that the
main controller encapsulates the control signal to a packet to
be sent, and t
rs
is the time instant that the system starts
processing the control signal. Fig. 2 shows the corresponding
timing diagram of network delay propagations.


Fig. 2 Timing diagram of network delay propagations

In fact, both network delays can be longer or shorter than
the sampling time T: The controller processing delay
c
and
both network delays can be lumped together as the control
delay for ease of analysis. This approach has been used in
some networked control methodologies. Although the
controller processing delay
c
always exists, this delay is
usually small compared to the network delays, and could be
neglected. In addition, the sampling periods of the main
controller and of the actuator may be different in some cases.
The delays
sc
and
ca
are composed of at least the following
parts [16]:
Waiting time delay
W
. The waiting time delay is the delay,
of which a source (the controller) has to wait for queuing
and network availability before actually sending a frame or a
packet out.
Frame time delay
F
. The frame time delay is the delay
during the moment that the source is placing a frame or a
packet on the network.
Propagation delay
P
. The propagation delay is the delay for
a frame or a packet traveling through a physical media. The
propagation delay depends on the speed of signal
transmission and the distance between the source and
destination.
These three delay parts are fundamental delays that occur
World Academy of Science, Engineering and Technology
Vol:2 2008-11-22
173
I
n
t
e
r
n
a
t
i
o
n
a
l

S
c
i
e
n
c
e

I
n
d
e
x

V
o
l
:
2
,

N
o
:
1
1
,

2
0
0
8

w
a
s
e
t
.
o
r
g
/
P
u
b
l
i
c
a
t
i
o
n
/
1
4
0
0
1



on a local area network. When the control or sensory data
travel across networks, there can be additional delays such as
the queuing delay at a switch or a router, and the propagation
delay between network hops. The delays
sc
and
ca
also
depend on other factors such as maximal bandwidths from
protocol specifications, and frame or packet sizes.
It is assumed that communication delays from the sensor to
controller and from controller to actuator are variables
(random), have the same values and they are uniform
distributed. The Simulink model for generation of these delays
is depicted in Fig. 3.

Out1
1
Zero-Order
Hold
Uniform random
number
Step1
Product

Fig. 3 System for generating a uniform distributed random signal

Communication delays which are larger then a sample time
(discretization delay) are considered in the paper. The sample
time is chosen to be 0.03 s. The communication time delay
generated with simulation scheme in Fig. 3, changed from 0 to
3*T (T is sample time), is shown in Fig. 4.
0 10 20 30 40 50
0
0.02
0.04
0.06
0.08
0.1
Time [s]
C
o
m
m
u
n
i
c
a
t
i
o
n

d
e
l
a
y

[
s
]

Fig. 4 Time distribution of communication delay under interval
[0,3T
d
]
B. Process Model and PI Controller
The PI controller is designed using technical optimum
technique. We used discrete model of PI controller described
by following equation:


i i
r
r
T z T
Tz K
z G

= ) ( , (2)

where K
r
is proportional gain and T
i
is integral time constant.
The process is represented by DC motor in the Laplace form:


1 7 . 0 1 . 0
1
) (
2
+ +
=
s s
s G . (3)

The process model, which predicts the future behavior of
process, is given in discrete form as follows:


2 1
2 1
43175 . 2 4072 . 5 3
011742 . 0 012593 . 0
) (


+
+
=
z z
z z
z G
m
. (4)

In the following section, the design of predictive controller
will be described.
III. DESIGN OF NETWORKED PREDICTIVE CONTROLLER
The proposed networked predictive controller contains the
ordinary controller, the process model, the adaptation of delay
loop and an appropriate filter. Process model and ordinary
controller have described in the previous section.
The adaptation algorithm is derived with assumption that
average communication delays from the sensor to the
controller and from the controller to the actuator are equal.
Both delays are variables and uniform distributed.
In adaptation loop, the communication delay in the
predictive controller is calculated based on N previous delays
in the network from the sensor to the controller:


N
d
N
i
sc
i

=
=
1
2

, (5)

where
sc
i
is communication delay from the sensor to the
controller in i-th step. Therefore, the average value of N
previous delays is calculated in this way.
For improving of robustness of the control system a first-
order filter is included in the feedback loop. The transfer
function of the filter is given by:


1
1
1
) (

=
z T
z G
f
f
, (6)

where T
f
is a time constant of the filter.
It is recommended to set T
f
as follows:

c
f
T
T = , (7)

where T
c
is a whole communication delay. The parameter
influences on the robustness of the control system. Smaller
values of provides better system robustness, while greater
value of it ensures faster compensation of disturbance
influence.
IV. SIMULATION RESULTS
Validations of proposed adaptive predictive controller in
comparison with direct and predictive controllers are
investigated on DC motor using Matlab/Simulink program. In
World Academy of Science, Engineering and Technology
Vol:2 2008-11-22
174
I
n
t
e
r
n
a
t
i
o
n
a
l

S
c
i
e
n
c
e

I
n
d
e
x

V
o
l
:
2
,

N
o
:
1
1
,

2
0
0
8

w
a
s
e
t
.
o
r
g
/
P
u
b
l
i
c
a
t
i
o
n
/
1
4
0
0
1



this section the influence of communication delays, sensor to
controller and controller and actuator, on control system
performance will be considered. In these simulations, we
assume that these delays are variables and take the same
values. The control performance of the closed-loop system for
three different communication delays is considered. Their
average values are equal to 3*T, 6*T and 9*T, respectively.
Two different input signals are applied to the system: step and
sequence of pulses. The parameters of adaptive predictive
controller are: K
r
=1.25, T
i
=0.5 s, T=0.03 s and T
f
=0.2 s.
Simulation results achieved by the step reference input are
shown in Figs. 5-8. In Fig. 5 the time responses of the direct
control system outputs (no prediction included) with and
without communication delays are presented. The average
delay is equal to 3*T. The time responses of the system with
predictive control, together with previous two are depicted in
Fig. 6. Simulation results with larger values of average
network communication delays: 6*T and 9*T are illustrated on
Figs. 7 and 8, respectively. The networked DC motor control
system with communication delay adaptation has superior
performance than without delay adaptation as indicated by the
lower overshoot.

0 1 2 3 4 5
0
0.2
0.4
0.6
0.8
1
1.2
Time [s]
V
o
l
t
a
g
e

[
V
]


without delay
with delay

Fig. 5 Step time responses of motor voltage with transportation delay
(average value is 3*T) and without delay under PI control



Fig. 6 Step time responses of motor voltage with PI control (with and
without time delay) and predictive control with time delay. Average
value of time delay is set to 3*T

0 1 2 3 4 5
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Time [s]
V
o
l
t
a
g
e

[
V
]


without delay
with delay
with predictor
with adaptive predictor

Fig. 7 Step time responses of motor voltage with PI control,
predictive control and predictive control with adaptation time delay.
Average value of time delay is 6*T
0 1 2 3 4 5
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
Time [s]
V
o
l
t
a
g
e

[
s
]



Fig. 8 Step time responses of motor voltage with PI control,
predictive control and predictive control with adaptation time delay.
Average value of time delay is 9*T

The simulation results of direct, predictive and adaptive
predictive controls, with the pulse sequence as an input signal,
are shown in Figs. 9-11. From these figures it can be
concluded that control performance by direct and predictive
controls with growth of delay are violated. In these cases, the
adaptive predictive controller demonstrates a good robustness
behavior. Consequently, the networked predictive control
scheme with adaptation loop can actively compensate for the
certain value of a communication (network) delay.
0 10 20 30 40 50
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
Time [s]
V
o
l
t
a
g
e

[
V
]

Fig. 9 Time responses of motor voltage on the pulse sequence: PI
control without delay (---) and with delay (), predictive control (
) and predictive control with adaptation delay (). Average value of
time delay is 3*T
0 1 2 3 4 5
0
0.
0.
0.
0.
1
1.
Time [s]
Voltage [V]


without delay
with delay
with delay and predictor
World Academy of Science, Engineering and Technology
Vol:2 2008-11-22
175
I
n
t
e
r
n
a
t
i
o
n
a
l

S
c
i
e
n
c
e

I
n
d
e
x

V
o
l
:
2
,

N
o
:
1
1
,

2
0
0
8

w
a
s
e
t
.
o
r
g
/
P
u
b
l
i
c
a
t
i
o
n
/
1
4
0
0
1



0 10 20 30 40 50
-0.5
0
0.5
1
1.5
Time [s]
V
o
l
t
a
g
e

[
V
]


Fig. 10 Time responses of motor voltage on the pulse sequence: PI
control without delay (---) and with delay (), predictive control (
) and predictive control with adaptation delay (). Average value of
time delay is 6*T

0 10 20 30 40 50
-0.5
0
0.5
1
1.5
Time [s]
V
o
l
t
a
g
e

[
V
]



Fig. 11 Time responses of motor voltage on the pulse sequence: PI
control without delay (---) and with delay (), predictive control (
) and predictive control with adaptation delay (). Average value of
time delay is 9*T
V. CONCLUSION
In this paper, the adaptive predictive controller is designed
to overcome the shortcomings of the influence of
communication delays among the nodes, such as controller,
sensor and actuator, in the network. The main component of
the proposed adaptive predictive controller is an adaptation
delay loop. This loop provides the satisfactory control
performance when the communication delay is greater then
sample time (even nine times). The simulation results obtained
show good performance of the proposed predictive controller
with an adaptation time delay loop.
REFERENCES
[1] V.J. VanDoren, The Smith Predictor: a Process Engineers Crystal
Ball, Control Engineering, May 1996.
[2] N.P. Mahalik, Fieldbus Technologies: Industrial Network Standards for
Real-Time Distributed Control. Berlin: Springer Verlag, 2005.
[3] R. Bannatyne, Microcontrollers for the automobile, Micro Control
Journal, 2003, http://www.mcjournal.com/articles/arc105/arc105.htm.
[4] S. Graham and P. R. Kumar, Time in general-purpose control systems:
The control time protocol and an experimental evaluation, in Proc. of
the 43rd IEEE Conference on Decision and Control, Dec 2004, pp.
40044009.
[5] Y. Tipsuwan, and M.Y. Chow, Control methodologies in networked
control systems, Control Engineering Practice, vol. 11, pp. 1099-1111,
2003.
[6] K. Shu, and J. Yan, "Robust Stability of Uncertain Time Delay Systems
and its Stabilization by Variable Structure Control," International
Journal of Control, vol. 57, no. 1, p237-246, 1993.
[7] G.P. Lui, and H. Wang, "Adaptive Controller for Continuous-Time
Systems with Unknown Varying Time Delay," 1991 Int. Conf. of Cont.
IEE Conf. Pub. v2, n332, p1084-1088.
[8] O.J.M. Smith, "A Controller to Overcome Dead Time" ISA Journal, No.
2,28, February 1959.
[9] Q. Brone, and S. Harris, "Varying Time Delay Estimation and Self-
Tuning Control," Proceedings form the 1991 American Controls
Conference, v2, p1740-1741.
[10] K. Passino and S. Yurkovich, Fuzzy Control, Addison- Wesley, 1998.
[11] Chan, H., & Ozguner, . U. (1995). Closed-loopcontrol of systems over a
communications network with queues. International Journal of Control,
62(3), 493510.
[12] Nilsson, J., 1998. Real-time control systems with delays. Ph.D.
dissertation, Lund Institute of Technology.
[13] G. C. Walsh, O. Beldiman, and L. Bushnell, Asymptotic behavior of
networked control systems, in Proc. of the 1999 IEEE international
conference on control applications, Kohala Coast, HI, vol. 2, 1999, pp.
14481453.
[14] S. H. Hong, and W. H. Kim, Bandwidth allocation scheme in CAN
protocol, IEE ProceedingsControl Theory and Applications, vol. 147,
no.1. pp. 3744, 2000.
[15] N. B. Almutairi, M.Y. Chow, and Y. Tipsuwan, Network-based
controlled DC motor with fuzzy compensation, in Proc. 27th annual
conference of the IEEE industrial electronics society (IECON 01),
Denver, CO, vol. 3, 2001, pp. 18441849.
[16] F. L. Lian, J. R. Moyne, and D. M. Tilbury, Performance evaluation of
control networks: Ethernet, Controlnet, and Device-Net, IEEE Control
Systems Magazine, vol. 21, no. 1, pp. 66-83, Feb. 2001.









World Academy of Science, Engineering and Technology
Vol:2 2008-11-22
176
I
n
t
e
r
n
a
t
i
o
n
a
l

S
c
i
e
n
c
e

I
n
d
e
x

V
o
l
:
2
,

N
o
:
1
1
,

2
0
0
8

w
a
s
e
t
.
o
r
g
/
P
u
b
l
i
c
a
t
i
o
n
/
1
4
0
0
1